Research On Vehicle Behavior Of Complex Weaving Areas Of Arterial In Mountainous Cities Based On Microscopic Trajectory Data

As an important part of urban trunk roads and expressway systems,the interweaving area often causes traffic congestion due to the complexity and turbulence of traffic flow.In order to accurately evaluate the capacity of the intertwined areas of mountainous urban roads,this paper based on the micro-trajectory data to study the driving behavior of urban urban trunk road interweaving area,and provide theoretical support for realizing urban fine control and improving the efficiency and safety level of urban trunk roads.Taking advantage of the advantages of UAV high-altitude video capture,this paper selects the complex interweaving areas of several typical mountain cities to study the vehicle behavior in the interwoven areas,mainly from the micro-trajectory data extraction based on UAV high-altitude video.The following aspects are studied: the mining of vehicle behavior characteristics in the Interleaver area,the construction and calibration of the VISSIM simulation model in the Interleaver area,and the optimization calculation model of the actual traffic capacity in the Interleaver area.The details as follows:(1)In order to realize the vehicle trajectory extraction of UAV video,a precise and efficient vehicle tracking algorithm is studied.In this paper,the video anti-shake processing is firstly based on the FAST feature extraction algorithm.Then,to solve the problem of target tracking loss caused by the complex video environment and target occlusion,the multi-scale KCF optimization algorithm is proposed in this paper.Compared with the Camshift algorithm and the original KCF algorithm,the experimental results show that the detection accuracy of the multi-scale KCF optimization algorithm in different scenarios is much higher than that of other algorithms.In the case of object occlusion and shadow,the average error is 3.32 pixels,which has strong robustness and applicability.(2)Vehicle micro trajectory data extraction based on UAV video.The vehicle trajectory data is extracted by using a multi-scale KCF optimization algorithm,including: headway spacing,speed,acceleration and etc.And the comparison error with the radar speed data is within ±4km/h,and the detection result has higher precision.(3)Vehicle feature behavior mining in the interweaving area,mainly from the aspects of microscopic vehicle behavior data and operational characteristics of the interweaving area.The micro trajectory data mainly includes: head spacing,speed,etc.;the operating characteristics of the interlaced area include: flow rate,interlace ratio of each lane,vehicle composition coefficient,and the like.According to the comparative analysis,the running states(flow,speed,and headway)of the lanes in the interlaced area corresponding to different traffic situations(interlaced flow ratio,remittance ratio,and interlacing ratio)are also different.(4)The traffic simulation model construction and parameter calibration method for the weaving areas of mountainous urban arterial are proposed.The measured data is used to calibrate the key parameters,and a reasonable threshold for each parameter is given.Finally,the four-kilometer interweaving area in Chongqing is used for comparative analysis.The experimental results show that the simulation results are within 7% compared with the actual flow rate in the case of congestion,which can accurately reflect the actual operation of the interweaving area.(5)Comparing the existing analysis method of interweaving area,it is found that the existing capacity calculation model is not applicable to the interweaving area of mountainous urban trunk roads,and the calculation result is much larger than the actual measurement capacity.Based on the VISSIM interweaving area simulation model to obtain the traffic capacity under different interleaving conditions,a traffic capacity calculation model suitable for the interweaving area of mountainous urban trunk roads is proposed.The length of the interlaced area(Ls),the number of interlaced areas(N),the interleaved flow ratio(QR)and The remittance ratio(QW)is used as the model's four parameters.The 1stOpt software is used to solve the nonlinear complex model parameters,The error with the measured data is abouSt 6%.